Computation Offloading Forh.264videoencoder on MobileDevices
Xiaoli Zhao,Shiqiang Yang
2006-01-01
Abstract:Multimedia application isaclass ofapplications thatwillbepopular onmobiledevices. Meantime, itisa kindofapplications thathascomplexcomputation, anda resource-killer application. Butresources, especially energy, areveryimportant formobile devices. Thereistheconflict between theresources limit andthecomputation complexity. Computation offloading isa methodtodealwiththis problem. Inthis paper, wemainly focus ontheusageofthe computation offloading methodtoH.264videoencoder on mobile devices. We propose H.264encoder modularization modelandenergymodels. Basedonthemodels, we also propose threeoffloading schemes. Thesimulation result of using theseoffloading schemes showsthatthecomputation offloading ofH.264videoencoderon mobiledevices is efficient andvaluable forenergy saving. Because ofsmall size, light weight andbattery-powered, mobile devices areveryconvenient tobeusedanywhere. Now,moreandmorepeople usemobile device, suchas mobile phone, Personal Digital Assistance (PDA), laptop computer andsoon.We canusethemtodealwiththe information anytime andanywhere. Besides this, more andmoremobile devices havewireless communication capabilities, so theusercangetinformation and communicate with moreconveniently. Togivetheuserbetter service, theengineers ofmobile devices develop moreandmoreapplications onmobile devices. Forsomeapplications, therequirement of resources maybeveryhigh, buttheresources onmobile devices areverylimited, suchasCPU,memory, energy andsoon.Amongthese resources, energy isthebiggest problem formobile devices. Somanytechnologies areproposed toprolong thebattery life onmobile devices. Someresearchers turnoffthe device thatisnotusedatthattime; someresearchers optimize theI/0device; andsomeresearchers slowdown theCPU.Allthese technologies focus onthehardware part. Computation offloading methodwhichusesthe wireless communication capability ofmobile devices isa software method todealwiththeproblem ofresources limit. Itoffloads somecomputations frommobile devices tothearound powerful server viawireless network. So thecomputation costofmobile devices decreases andthe pressure ofresources usageonmobile devices canbe relaxed. Thevideo processing application isakindofpopular applications onmobile devices. Someone mayusevideo decoder toplayavideosequence. Someonemayuse video encoder tomakethevideo havesmall size, sothe video isveryconvenient tosaveonmobile devices or sendtofriends faraway.Butthis kindofapplication, especially videoencoderapplication, needshuge computation resources. So we applycomputation offloading forvideo encoder. Nowthere aremanykinds ofvideo encoders based ondifferent encoder standards, suchasMpeg2,Mpeg4,H.263, H.264/AVC andsoon. H.264/AVC videoencoder hasmoreadvantages than otherstandards, forexample itcan gethigher compression. Sowethink itwill bepopular onmobile devices, butitalso hashigh resource requirement, sowe focus onH.264/AVC video encoder application inthis paperandapply computation offloading method toitfor saving energy resource onmobile devices. Theremaining sections ofthepaper areorganized inthe following order. Insection II,we briefly survey the related workofcomputation offloading technology. In section III, we present thecomputation offloading technology andtheconditions whenweuseit. Insection IV,we present themodularization ofH.264video encoder andenergymodels. Thenwe do some experiments andanalyze theoffloading ofH.264video encoder insection V.Finally, weconclude insection VI.